xref: /dflybsd-src/sys/kern/sys_pipe.c (revision dcaa8a41662f2b0cf579a6e912564c9fc8275ac1)
1 /*
2  * Copyright (c) 1996 John S. Dyson
3  * All rights reserved.
4  *
5  * Redistribution and use in source and binary forms, with or without
6  * modification, are permitted provided that the following conditions
7  * are met:
8  * 1. Redistributions of source code must retain the above copyright
9  *    notice immediately at the beginning of the file, without modification,
10  *    this list of conditions, and the following disclaimer.
11  * 2. Redistributions in binary form must reproduce the above copyright
12  *    notice, this list of conditions and the following disclaimer in the
13  *    documentation and/or other materials provided with the distribution.
14  * 3. Absolutely no warranty of function or purpose is made by the author
15  *    John S. Dyson.
16  * 4. Modifications may be freely made to this file if the above conditions
17  *    are met.
18  *
19  * $FreeBSD: src/sys/kern/sys_pipe.c,v 1.60.2.13 2002/08/05 15:05:15 des Exp $
20  * $DragonFly: src/sys/kern/sys_pipe.c,v 1.50 2008/09/09 04:06:13 dillon Exp $
21  */
22 
23 /*
24  * This file contains a high-performance replacement for the socket-based
25  * pipes scheme originally used in FreeBSD/4.4Lite.  It does not support
26  * all features of sockets, but does do everything that pipes normally
27  * do.
28  */
29 #include <sys/param.h>
30 #include <sys/systm.h>
31 #include <sys/kernel.h>
32 #include <sys/proc.h>
33 #include <sys/fcntl.h>
34 #include <sys/file.h>
35 #include <sys/filedesc.h>
36 #include <sys/filio.h>
37 #include <sys/ttycom.h>
38 #include <sys/stat.h>
39 #include <sys/signalvar.h>
40 #include <sys/sysproto.h>
41 #include <sys/pipe.h>
42 #include <sys/vnode.h>
43 #include <sys/uio.h>
44 #include <sys/event.h>
45 #include <sys/globaldata.h>
46 #include <sys/module.h>
47 #include <sys/malloc.h>
48 #include <sys/sysctl.h>
49 #include <sys/socket.h>
50 
51 #include <vm/vm.h>
52 #include <vm/vm_param.h>
53 #include <sys/lock.h>
54 #include <vm/vm_object.h>
55 #include <vm/vm_kern.h>
56 #include <vm/vm_extern.h>
57 #include <vm/pmap.h>
58 #include <vm/vm_map.h>
59 #include <vm/vm_page.h>
60 #include <vm/vm_zone.h>
61 
62 #include <sys/file2.h>
63 #include <sys/signal2.h>
64 #include <sys/mplock2.h>
65 
66 #include <machine/cpufunc.h>
67 
68 /*
69  * interfaces to the outside world
70  */
71 static int pipe_read (struct file *fp, struct uio *uio,
72 		struct ucred *cred, int flags);
73 static int pipe_write (struct file *fp, struct uio *uio,
74 		struct ucred *cred, int flags);
75 static int pipe_close (struct file *fp);
76 static int pipe_shutdown (struct file *fp, int how);
77 static int pipe_kqfilter (struct file *fp, struct knote *kn);
78 static int pipe_stat (struct file *fp, struct stat *sb, struct ucred *cred);
79 static int pipe_ioctl (struct file *fp, u_long cmd, caddr_t data,
80 		struct ucred *cred, struct sysmsg *msg);
81 
82 static struct fileops pipeops = {
83 	.fo_read = pipe_read,
84 	.fo_write = pipe_write,
85 	.fo_ioctl = pipe_ioctl,
86 	.fo_kqfilter = pipe_kqfilter,
87 	.fo_stat = pipe_stat,
88 	.fo_close = pipe_close,
89 	.fo_shutdown = pipe_shutdown
90 };
91 
92 static void	filt_pipedetach(struct knote *kn);
93 static int	filt_piperead(struct knote *kn, long hint);
94 static int	filt_pipewrite(struct knote *kn, long hint);
95 
96 static struct filterops pipe_rfiltops =
97 	{ FILTEROP_ISFD, NULL, filt_pipedetach, filt_piperead };
98 static struct filterops pipe_wfiltops =
99 	{ FILTEROP_ISFD, NULL, filt_pipedetach, filt_pipewrite };
100 
101 MALLOC_DEFINE(M_PIPE, "pipe", "pipe structures");
102 
103 /*
104  * Default pipe buffer size(s), this can be kind-of large now because pipe
105  * space is pageable.  The pipe code will try to maintain locality of
106  * reference for performance reasons, so small amounts of outstanding I/O
107  * will not wipe the cache.
108  */
109 #define MINPIPESIZE (PIPE_SIZE/3)
110 #define MAXPIPESIZE (2*PIPE_SIZE/3)
111 
112 /*
113  * Limit the number of "big" pipes
114  */
115 #define LIMITBIGPIPES	64
116 #define PIPEQ_MAX_CACHE 16      /* per-cpu pipe structure cache */
117 
118 static int pipe_maxbig = LIMITBIGPIPES;
119 static int pipe_maxcache = PIPEQ_MAX_CACHE;
120 static int pipe_bigcount;
121 static int pipe_nbig;
122 static int pipe_bcache_alloc;
123 static int pipe_bkmem_alloc;
124 static int pipe_rblocked_count;
125 static int pipe_wblocked_count;
126 
127 SYSCTL_NODE(_kern, OID_AUTO, pipe, CTLFLAG_RW, 0, "Pipe operation");
128 SYSCTL_INT(_kern_pipe, OID_AUTO, nbig,
129         CTLFLAG_RD, &pipe_nbig, 0, "numer of big pipes allocated");
130 SYSCTL_INT(_kern_pipe, OID_AUTO, bigcount,
131         CTLFLAG_RW, &pipe_bigcount, 0, "number of times pipe expanded");
132 SYSCTL_INT(_kern_pipe, OID_AUTO, rblocked,
133         CTLFLAG_RW, &pipe_rblocked_count, 0, "number of times pipe expanded");
134 SYSCTL_INT(_kern_pipe, OID_AUTO, wblocked,
135         CTLFLAG_RW, &pipe_wblocked_count, 0, "number of times pipe expanded");
136 SYSCTL_INT(_kern_pipe, OID_AUTO, maxcache,
137         CTLFLAG_RW, &pipe_maxcache, 0, "max pipes cached per-cpu");
138 SYSCTL_INT(_kern_pipe, OID_AUTO, maxbig,
139         CTLFLAG_RW, &pipe_maxbig, 0, "max number of big pipes");
140 #ifdef SMP
141 static int pipe_delay = 5000;	/* 5uS default */
142 SYSCTL_INT(_kern_pipe, OID_AUTO, delay,
143         CTLFLAG_RW, &pipe_delay, 0, "SMP delay optimization in ns");
144 static int pipe_mpsafe = 1;
145 SYSCTL_INT(_kern_pipe, OID_AUTO, mpsafe,
146         CTLFLAG_RW, &pipe_mpsafe, 0, "");
147 #endif
148 #if !defined(NO_PIPE_SYSCTL_STATS)
149 SYSCTL_INT(_kern_pipe, OID_AUTO, bcache_alloc,
150         CTLFLAG_RW, &pipe_bcache_alloc, 0, "pipe buffer from pcpu cache");
151 SYSCTL_INT(_kern_pipe, OID_AUTO, bkmem_alloc,
152         CTLFLAG_RW, &pipe_bkmem_alloc, 0, "pipe buffer from kmem");
153 #endif
154 
155 static void pipeclose (struct pipe *cpipe);
156 static void pipe_free_kmem (struct pipe *cpipe);
157 static int pipe_create (struct pipe **cpipep);
158 static int pipespace (struct pipe *cpipe, int size);
159 
160 static __inline void
161 pipewakeup(struct pipe *cpipe, int dosigio)
162 {
163 	if (dosigio && (cpipe->pipe_state & PIPE_ASYNC) && cpipe->pipe_sigio) {
164 		get_mplock();
165 		pgsigio(cpipe->pipe_sigio, SIGIO, 0);
166 		rel_mplock();
167 	}
168 	KNOTE(&cpipe->pipe_kq.ki_note, 0);
169 }
170 
171 /*
172  * These routines are called before and after a UIO.  The UIO
173  * may block, causing our held tokens to be lost temporarily.
174  *
175  * We use these routines to serialize reads against other reads
176  * and writes against other writes.
177  *
178  * The read token is held on entry so *ipp does not race.
179  */
180 static __inline int
181 pipe_start_uio(struct pipe *cpipe, int *ipp)
182 {
183 	int error;
184 
185 	while (*ipp) {
186 		*ipp = -1;
187 		error = tsleep(ipp, PCATCH, "pipexx", 0);
188 		if (error)
189 			return (error);
190 	}
191 	*ipp = 1;
192 	return (0);
193 }
194 
195 static __inline void
196 pipe_end_uio(struct pipe *cpipe, int *ipp)
197 {
198 	if (*ipp < 0) {
199 		*ipp = 0;
200 		wakeup(ipp);
201 	} else {
202 		KKASSERT(*ipp > 0);
203 		*ipp = 0;
204 	}
205 }
206 
207 static __inline void
208 pipe_get_mplock(int *save)
209 {
210 #ifdef SMP
211 	if (pipe_mpsafe == 0) {
212 		get_mplock();
213 		*save = 1;
214 	} else
215 #endif
216 	{
217 		*save = 0;
218 	}
219 }
220 
221 static __inline void
222 pipe_rel_mplock(int *save)
223 {
224 #ifdef SMP
225 	if (*save)
226 		rel_mplock();
227 #endif
228 }
229 
230 
231 /*
232  * The pipe system call for the DTYPE_PIPE type of pipes
233  *
234  * pipe_args(int dummy)
235  *
236  * MPSAFE
237  */
238 int
239 sys_pipe(struct pipe_args *uap)
240 {
241 	struct thread *td = curthread;
242 	struct filedesc *fdp = td->td_proc->p_fd;
243 	struct file *rf, *wf;
244 	struct pipe *rpipe, *wpipe;
245 	int fd1, fd2, error;
246 
247 	rpipe = wpipe = NULL;
248 	if (pipe_create(&rpipe) || pipe_create(&wpipe)) {
249 		pipeclose(rpipe);
250 		pipeclose(wpipe);
251 		return (ENFILE);
252 	}
253 
254 	error = falloc(td->td_lwp, &rf, &fd1);
255 	if (error) {
256 		pipeclose(rpipe);
257 		pipeclose(wpipe);
258 		return (error);
259 	}
260 	uap->sysmsg_fds[0] = fd1;
261 
262 	/*
263 	 * Warning: once we've gotten past allocation of the fd for the
264 	 * read-side, we can only drop the read side via fdrop() in order
265 	 * to avoid races against processes which manage to dup() the read
266 	 * side while we are blocked trying to allocate the write side.
267 	 */
268 	rf->f_type = DTYPE_PIPE;
269 	rf->f_flag = FREAD | FWRITE;
270 	rf->f_ops = &pipeops;
271 	rf->f_data = rpipe;
272 	error = falloc(td->td_lwp, &wf, &fd2);
273 	if (error) {
274 		fsetfd(fdp, NULL, fd1);
275 		fdrop(rf);
276 		/* rpipe has been closed by fdrop(). */
277 		pipeclose(wpipe);
278 		return (error);
279 	}
280 	wf->f_type = DTYPE_PIPE;
281 	wf->f_flag = FREAD | FWRITE;
282 	wf->f_ops = &pipeops;
283 	wf->f_data = wpipe;
284 	uap->sysmsg_fds[1] = fd2;
285 
286 	rpipe->pipe_slock = kmalloc(sizeof(struct lock),
287 				    M_PIPE, M_WAITOK|M_ZERO);
288 	wpipe->pipe_slock = rpipe->pipe_slock;
289 	rpipe->pipe_peer = wpipe;
290 	wpipe->pipe_peer = rpipe;
291 	lockinit(rpipe->pipe_slock, "pipecl", 0, 0);
292 
293 	/*
294 	 * Once activated the peer relationship remains valid until
295 	 * both sides are closed.
296 	 */
297 	fsetfd(fdp, rf, fd1);
298 	fsetfd(fdp, wf, fd2);
299 	fdrop(rf);
300 	fdrop(wf);
301 
302 	return (0);
303 }
304 
305 /*
306  * Allocate kva for pipe circular buffer, the space is pageable
307  * This routine will 'realloc' the size of a pipe safely, if it fails
308  * it will retain the old buffer.
309  * If it fails it will return ENOMEM.
310  */
311 static int
312 pipespace(struct pipe *cpipe, int size)
313 {
314 	struct vm_object *object;
315 	caddr_t buffer;
316 	int npages, error;
317 
318 	npages = round_page(size) / PAGE_SIZE;
319 	object = cpipe->pipe_buffer.object;
320 
321 	/*
322 	 * [re]create the object if necessary and reserve space for it
323 	 * in the kernel_map.  The object and memory are pageable.  On
324 	 * success, free the old resources before assigning the new
325 	 * ones.
326 	 */
327 	if (object == NULL || object->size != npages) {
328 		get_mplock();
329 		object = vm_object_allocate(OBJT_DEFAULT, npages);
330 		buffer = (caddr_t)vm_map_min(&kernel_map);
331 
332 		error = vm_map_find(&kernel_map, object, 0,
333 				    (vm_offset_t *)&buffer,
334 				    size, PAGE_SIZE,
335 				    1, VM_MAPTYPE_NORMAL,
336 				    VM_PROT_ALL, VM_PROT_ALL,
337 				    0);
338 
339 		if (error != KERN_SUCCESS) {
340 			vm_object_deallocate(object);
341 			rel_mplock();
342 			return (ENOMEM);
343 		}
344 		pipe_free_kmem(cpipe);
345 		rel_mplock();
346 		cpipe->pipe_buffer.object = object;
347 		cpipe->pipe_buffer.buffer = buffer;
348 		cpipe->pipe_buffer.size = size;
349 		++pipe_bkmem_alloc;
350 	} else {
351 		++pipe_bcache_alloc;
352 	}
353 	cpipe->pipe_buffer.rindex = 0;
354 	cpipe->pipe_buffer.windex = 0;
355 	return (0);
356 }
357 
358 /*
359  * Initialize and allocate VM and memory for pipe, pulling the pipe from
360  * our per-cpu cache if possible.  For now make sure it is sized for the
361  * smaller PIPE_SIZE default.
362  */
363 static int
364 pipe_create(struct pipe **cpipep)
365 {
366 	globaldata_t gd = mycpu;
367 	struct pipe *cpipe;
368 	int error;
369 
370 	if ((cpipe = gd->gd_pipeq) != NULL) {
371 		gd->gd_pipeq = cpipe->pipe_peer;
372 		--gd->gd_pipeqcount;
373 		cpipe->pipe_peer = NULL;
374 		cpipe->pipe_wantwcnt = 0;
375 	} else {
376 		cpipe = kmalloc(sizeof(struct pipe), M_PIPE, M_WAITOK|M_ZERO);
377 	}
378 	*cpipep = cpipe;
379 	if ((error = pipespace(cpipe, PIPE_SIZE)) != 0)
380 		return (error);
381 	vfs_timestamp(&cpipe->pipe_ctime);
382 	cpipe->pipe_atime = cpipe->pipe_ctime;
383 	cpipe->pipe_mtime = cpipe->pipe_ctime;
384 	lwkt_token_init(&cpipe->pipe_rlock, 1, "piper");
385 	lwkt_token_init(&cpipe->pipe_wlock, 1, "pipew");
386 	return (0);
387 }
388 
389 /*
390  * MPALMOSTSAFE (acquires mplock)
391  */
392 static int
393 pipe_read(struct file *fp, struct uio *uio, struct ucred *cred, int fflags)
394 {
395 	struct pipe *rpipe;
396 	struct pipe *wpipe;
397 	int error;
398 	size_t nread = 0;
399 	int nbio;
400 	u_int size;	/* total bytes available */
401 	u_int nsize;	/* total bytes to read */
402 	u_int rindex;	/* contiguous bytes available */
403 	int notify_writer;
404 	int mpsave;
405 	int bigread;
406 	int bigcount;
407 
408 	if (uio->uio_resid == 0)
409 		return(0);
410 
411 	/*
412 	 * Setup locks, calculate nbio
413 	 */
414 	pipe_get_mplock(&mpsave);
415 	rpipe = (struct pipe *)fp->f_data;
416 	wpipe = rpipe->pipe_peer;
417 	lwkt_gettoken(&rpipe->pipe_rlock);
418 
419 	if (fflags & O_FBLOCKING)
420 		nbio = 0;
421 	else if (fflags & O_FNONBLOCKING)
422 		nbio = 1;
423 	else if (fp->f_flag & O_NONBLOCK)
424 		nbio = 1;
425 	else
426 		nbio = 0;
427 
428 	/*
429 	 * Reads are serialized.  Note however that pipe_buffer.buffer and
430 	 * pipe_buffer.size can change out from under us when the number
431 	 * of bytes in the buffer are zero due to the write-side doing a
432 	 * pipespace().
433 	 */
434 	error = pipe_start_uio(rpipe, &rpipe->pipe_rip);
435 	if (error) {
436 		pipe_rel_mplock(&mpsave);
437 		lwkt_reltoken(&rpipe->pipe_rlock);
438 		return (error);
439 	}
440 	notify_writer = 0;
441 
442 	bigread = (uio->uio_resid > 10 * 1024 * 1024);
443 	bigcount = 10;
444 
445 	while (uio->uio_resid) {
446 		/*
447 		 * Don't hog the cpu.
448 		 */
449 		if (bigread && --bigcount == 0) {
450 			lwkt_user_yield();
451 			bigcount = 10;
452 			if (CURSIG(curthread->td_lwp)) {
453 				error = EINTR;
454 				break;
455 			}
456 		}
457 
458 		size = rpipe->pipe_buffer.windex - rpipe->pipe_buffer.rindex;
459 		cpu_lfence();
460 		if (size) {
461 			rindex = rpipe->pipe_buffer.rindex &
462 				 (rpipe->pipe_buffer.size - 1);
463 			nsize = size;
464 			if (nsize > rpipe->pipe_buffer.size - rindex)
465 				nsize = rpipe->pipe_buffer.size - rindex;
466 			nsize = szmin(nsize, uio->uio_resid);
467 
468 			error = uiomove(&rpipe->pipe_buffer.buffer[rindex],
469 					nsize, uio);
470 			if (error)
471 				break;
472 			cpu_mfence();
473 			rpipe->pipe_buffer.rindex += nsize;
474 			nread += nsize;
475 
476 			/*
477 			 * If the FIFO is still over half full just continue
478 			 * and do not try to notify the writer yet.
479 			 */
480 			if (size - nsize >= (rpipe->pipe_buffer.size >> 1)) {
481 				notify_writer = 0;
482 				continue;
483 			}
484 
485 			/*
486 			 * When the FIFO is less then half full notify any
487 			 * waiting writer.  WANTW can be checked while
488 			 * holding just the rlock.
489 			 */
490 			notify_writer = 1;
491 			if ((rpipe->pipe_state & PIPE_WANTW) == 0)
492 				continue;
493 		}
494 
495 		/*
496 		 * If the "write-side" was blocked we wake it up.  This code
497 		 * is reached either when the buffer is completely emptied
498 		 * or if it becomes more then half-empty.
499 		 *
500 		 * Pipe_state can only be modified if both the rlock and
501 		 * wlock are held.
502 		 */
503 		if (rpipe->pipe_state & PIPE_WANTW) {
504 			lwkt_gettoken(&rpipe->pipe_wlock);
505 			if (rpipe->pipe_state & PIPE_WANTW) {
506 				rpipe->pipe_state &= ~PIPE_WANTW;
507 				lwkt_reltoken(&rpipe->pipe_wlock);
508 				wakeup(rpipe);
509 			} else {
510 				lwkt_reltoken(&rpipe->pipe_wlock);
511 			}
512 		}
513 
514 		/*
515 		 * Pick up our copy loop again if the writer sent data to
516 		 * us while we were messing around.
517 		 *
518 		 * On a SMP box poll up to pipe_delay nanoseconds for new
519 		 * data.  Typically a value of 2000 to 4000 is sufficient
520 		 * to eradicate most IPIs/tsleeps/wakeups when a pipe
521 		 * is used for synchronous communications with small packets,
522 		 * and 8000 or so (8uS) will pipeline large buffer xfers
523 		 * between cpus over a pipe.
524 		 *
525 		 * For synchronous communications a hit means doing a
526 		 * full Awrite-Bread-Bwrite-Aread cycle in less then 2uS,
527 		 * where as miss requiring a tsleep/wakeup sequence
528 		 * will take 7uS or more.
529 		 */
530 		if (rpipe->pipe_buffer.windex != rpipe->pipe_buffer.rindex)
531 			continue;
532 
533 #if defined(SMP) && defined(_RDTSC_SUPPORTED_)
534 		if (pipe_delay) {
535 			int64_t tsc_target;
536 			int good = 0;
537 
538 			tsc_target = tsc_get_target(pipe_delay);
539 			while (tsc_test_target(tsc_target) == 0) {
540 				if (rpipe->pipe_buffer.windex !=
541 				    rpipe->pipe_buffer.rindex) {
542 					good = 1;
543 					break;
544 				}
545 			}
546 			if (good)
547 				continue;
548 		}
549 #endif
550 
551 		/*
552 		 * Detect EOF condition, do not set error.
553 		 */
554 		if (rpipe->pipe_state & PIPE_REOF)
555 			break;
556 
557 		/*
558 		 * Break if some data was read, or if this was a non-blocking
559 		 * read.
560 		 */
561 		if (nread > 0)
562 			break;
563 
564 		if (nbio) {
565 			error = EAGAIN;
566 			break;
567 		}
568 
569 		/*
570 		 * Last chance, interlock with WANTR.
571 		 */
572 		lwkt_gettoken(&rpipe->pipe_wlock);
573 		size = rpipe->pipe_buffer.windex - rpipe->pipe_buffer.rindex;
574 		if (size) {
575 			lwkt_reltoken(&rpipe->pipe_wlock);
576 			continue;
577 		}
578 
579 		/*
580 		 * Retest EOF - acquiring a new token can temporarily release
581 		 * tokens already held.
582 		 */
583 		if (rpipe->pipe_state & PIPE_REOF) {
584 			lwkt_reltoken(&rpipe->pipe_wlock);
585 			break;
586 		}
587 
588 		/*
589 		 * If there is no more to read in the pipe, reset its
590 		 * pointers to the beginning.  This improves cache hit
591 		 * stats.
592 		 *
593 		 * We need both locks to modify both pointers, and there
594 		 * must also not be a write in progress or the uiomove()
595 		 * in the write might block and temporarily release
596 		 * its wlock, then reacquire and update windex.  We are
597 		 * only serialized against reads, not writes.
598 		 *
599 		 * XXX should we even bother resetting the indices?  It
600 		 *     might actually be more cache efficient not to.
601 		 */
602 		if (rpipe->pipe_buffer.rindex == rpipe->pipe_buffer.windex &&
603 		    rpipe->pipe_wip == 0) {
604 			rpipe->pipe_buffer.rindex = 0;
605 			rpipe->pipe_buffer.windex = 0;
606 		}
607 
608 		/*
609 		 * Wait for more data.
610 		 *
611 		 * Pipe_state can only be set if both the rlock and wlock
612 		 * are held.
613 		 */
614 		rpipe->pipe_state |= PIPE_WANTR;
615 		tsleep_interlock(rpipe, PCATCH);
616 		lwkt_reltoken(&rpipe->pipe_wlock);
617 		error = tsleep(rpipe, PCATCH | PINTERLOCKED, "piperd", 0);
618 		++pipe_rblocked_count;
619 		if (error)
620 			break;
621 	}
622 	pipe_end_uio(rpipe, &rpipe->pipe_rip);
623 
624 	/*
625 	 * Uptime last access time
626 	 */
627 	if (error == 0 && nread)
628 		vfs_timestamp(&rpipe->pipe_atime);
629 
630 	/*
631 	 * If we drained the FIFO more then half way then handle
632 	 * write blocking hysteresis.
633 	 *
634 	 * Note that PIPE_WANTW cannot be set by the writer without
635 	 * it holding both rlock and wlock, so we can test it
636 	 * while holding just rlock.
637 	 */
638 	if (notify_writer) {
639 		/*
640 		 * Synchronous blocking is done on the pipe involved
641 		 */
642 		if (rpipe->pipe_state & PIPE_WANTW) {
643 			lwkt_gettoken(&rpipe->pipe_wlock);
644 			if (rpipe->pipe_state & PIPE_WANTW) {
645 				rpipe->pipe_state &= ~PIPE_WANTW;
646 				lwkt_reltoken(&rpipe->pipe_wlock);
647 				wakeup(rpipe);
648 			} else {
649 				lwkt_reltoken(&rpipe->pipe_wlock);
650 			}
651 		}
652 
653 		/*
654 		 * But we may also have to deal with a kqueue which is
655 		 * stored on the same pipe as its descriptor, so a
656 		 * EVFILT_WRITE event waiting for our side to drain will
657 		 * be on the other side.
658 		 */
659 		lwkt_gettoken(&wpipe->pipe_wlock);
660 		pipewakeup(wpipe, 0);
661 		lwkt_reltoken(&wpipe->pipe_wlock);
662 	}
663 	/*size = rpipe->pipe_buffer.windex - rpipe->pipe_buffer.rindex;*/
664 	lwkt_reltoken(&rpipe->pipe_rlock);
665 
666 	pipe_rel_mplock(&mpsave);
667 	return (error);
668 }
669 
670 /*
671  * MPALMOSTSAFE - acquires mplock
672  */
673 static int
674 pipe_write(struct file *fp, struct uio *uio, struct ucred *cred, int fflags)
675 {
676 	int error;
677 	int orig_resid;
678 	int nbio;
679 	struct pipe *wpipe;
680 	struct pipe *rpipe;
681 	u_int windex;
682 	u_int space;
683 	u_int wcount;
684 	int mpsave;
685 	int bigwrite;
686 	int bigcount;
687 
688 	pipe_get_mplock(&mpsave);
689 
690 	/*
691 	 * Writes go to the peer.  The peer will always exist.
692 	 */
693 	rpipe = (struct pipe *) fp->f_data;
694 	wpipe = rpipe->pipe_peer;
695 	lwkt_gettoken(&wpipe->pipe_wlock);
696 	if (wpipe->pipe_state & PIPE_WEOF) {
697 		pipe_rel_mplock(&mpsave);
698 		lwkt_reltoken(&wpipe->pipe_wlock);
699 		return (EPIPE);
700 	}
701 
702 	/*
703 	 * Degenerate case (EPIPE takes prec)
704 	 */
705 	if (uio->uio_resid == 0) {
706 		pipe_rel_mplock(&mpsave);
707 		lwkt_reltoken(&wpipe->pipe_wlock);
708 		return(0);
709 	}
710 
711 	/*
712 	 * Writes are serialized (start_uio must be called with wlock)
713 	 */
714 	error = pipe_start_uio(wpipe, &wpipe->pipe_wip);
715 	if (error) {
716 		pipe_rel_mplock(&mpsave);
717 		lwkt_reltoken(&wpipe->pipe_wlock);
718 		return (error);
719 	}
720 
721 	if (fflags & O_FBLOCKING)
722 		nbio = 0;
723 	else if (fflags & O_FNONBLOCKING)
724 		nbio = 1;
725 	else if (fp->f_flag & O_NONBLOCK)
726 		nbio = 1;
727 	else
728 		nbio = 0;
729 
730 	/*
731 	 * If it is advantageous to resize the pipe buffer, do
732 	 * so.  We are write-serialized so we can block safely.
733 	 */
734 	if ((wpipe->pipe_buffer.size <= PIPE_SIZE) &&
735 	    (pipe_nbig < pipe_maxbig) &&
736 	    wpipe->pipe_wantwcnt > 4 &&
737 	    (wpipe->pipe_buffer.rindex == wpipe->pipe_buffer.windex)) {
738 		/*
739 		 * Recheck after lock.
740 		 */
741 		lwkt_gettoken(&wpipe->pipe_rlock);
742 		if ((wpipe->pipe_buffer.size <= PIPE_SIZE) &&
743 		    (pipe_nbig < pipe_maxbig) &&
744 		    (wpipe->pipe_buffer.rindex == wpipe->pipe_buffer.windex)) {
745 			atomic_add_int(&pipe_nbig, 1);
746 			if (pipespace(wpipe, BIG_PIPE_SIZE) == 0)
747 				++pipe_bigcount;
748 			else
749 				atomic_subtract_int(&pipe_nbig, 1);
750 		}
751 		lwkt_reltoken(&wpipe->pipe_rlock);
752 	}
753 
754 	orig_resid = uio->uio_resid;
755 	wcount = 0;
756 
757 	bigwrite = (uio->uio_resid > 10 * 1024 * 1024);
758 	bigcount = 10;
759 
760 	while (uio->uio_resid) {
761 		if (wpipe->pipe_state & PIPE_WEOF) {
762 			error = EPIPE;
763 			break;
764 		}
765 
766 		/*
767 		 * Don't hog the cpu.
768 		 */
769 		if (bigwrite && --bigcount == 0) {
770 			lwkt_user_yield();
771 			bigcount = 10;
772 			if (CURSIG(curthread->td_lwp)) {
773 				error = EINTR;
774 				break;
775 			}
776 		}
777 
778 		windex = wpipe->pipe_buffer.windex &
779 			 (wpipe->pipe_buffer.size - 1);
780 		space = wpipe->pipe_buffer.size -
781 			(wpipe->pipe_buffer.windex - wpipe->pipe_buffer.rindex);
782 		cpu_lfence();
783 
784 		/* Writes of size <= PIPE_BUF must be atomic. */
785 		if ((space < uio->uio_resid) && (orig_resid <= PIPE_BUF))
786 			space = 0;
787 
788 		/*
789 		 * Write to fill, read size handles write hysteresis.  Also
790 		 * additional restrictions can cause select-based non-blocking
791 		 * writes to spin.
792 		 */
793 		if (space > 0) {
794 			u_int segsize;
795 
796 			/*
797 			 * Transfer size is minimum of uio transfer
798 			 * and free space in pipe buffer.
799 			 *
800 			 * Limit each uiocopy to no more then PIPE_SIZE
801 			 * so we can keep the gravy train going on a
802 			 * SMP box.  This doubles the performance for
803 			 * write sizes > 16K.  Otherwise large writes
804 			 * wind up doing an inefficient synchronous
805 			 * ping-pong.
806 			 */
807 			space = szmin(space, uio->uio_resid);
808 			if (space > PIPE_SIZE)
809 				space = PIPE_SIZE;
810 
811 			/*
812 			 * First segment to transfer is minimum of
813 			 * transfer size and contiguous space in
814 			 * pipe buffer.  If first segment to transfer
815 			 * is less than the transfer size, we've got
816 			 * a wraparound in the buffer.
817 			 */
818 			segsize = wpipe->pipe_buffer.size - windex;
819 			if (segsize > space)
820 				segsize = space;
821 
822 #ifdef SMP
823 			/*
824 			 * If this is the first loop and the reader is
825 			 * blocked, do a preemptive wakeup of the reader.
826 			 *
827 			 * On SMP the IPI latency plus the wlock interlock
828 			 * on the reader side is the fastest way to get the
829 			 * reader going.  (The scheduler will hard loop on
830 			 * lock tokens).
831 			 *
832 			 * NOTE: We can't clear WANTR here without acquiring
833 			 * the rlock, which we don't want to do here!
834 			 */
835 			if ((wpipe->pipe_state & PIPE_WANTR) && pipe_mpsafe > 1)
836 				wakeup(wpipe);
837 #endif
838 
839 			/*
840 			 * Transfer segment, which may include a wrap-around.
841 			 * Update windex to account for both all in one go
842 			 * so the reader can read() the data atomically.
843 			 */
844 			error = uiomove(&wpipe->pipe_buffer.buffer[windex],
845 					segsize, uio);
846 			if (error == 0 && segsize < space) {
847 				segsize = space - segsize;
848 				error = uiomove(&wpipe->pipe_buffer.buffer[0],
849 						segsize, uio);
850 			}
851 			if (error)
852 				break;
853 			cpu_mfence();
854 			wpipe->pipe_buffer.windex += space;
855 			wcount += space;
856 			continue;
857 		}
858 
859 		/*
860 		 * We need both the rlock and the wlock to interlock against
861 		 * the EOF, WANTW, and size checks, and to modify pipe_state.
862 		 *
863 		 * These are token locks so we do not have to worry about
864 		 * deadlocks.
865 		 */
866 		lwkt_gettoken(&wpipe->pipe_rlock);
867 
868 		/*
869 		 * If the "read-side" has been blocked, wake it up now
870 		 * and yield to let it drain synchronously rather
871 		 * then block.
872 		 */
873 		if (wpipe->pipe_state & PIPE_WANTR) {
874 			wpipe->pipe_state &= ~PIPE_WANTR;
875 			wakeup(wpipe);
876 		}
877 
878 		/*
879 		 * don't block on non-blocking I/O
880 		 */
881 		if (nbio) {
882 			lwkt_reltoken(&wpipe->pipe_rlock);
883 			error = EAGAIN;
884 			break;
885 		}
886 
887 		/*
888 		 * re-test whether we have to block in the writer after
889 		 * acquiring both locks, in case the reader opened up
890 		 * some space.
891 		 */
892 		space = wpipe->pipe_buffer.size -
893 			(wpipe->pipe_buffer.windex - wpipe->pipe_buffer.rindex);
894 		cpu_lfence();
895 		if ((space < uio->uio_resid) && (orig_resid <= PIPE_BUF))
896 			space = 0;
897 
898 		/*
899 		 * Retest EOF - acquiring a new token can temporarily release
900 		 * tokens already held.
901 		 */
902 		if (wpipe->pipe_state & PIPE_WEOF) {
903 			lwkt_reltoken(&wpipe->pipe_rlock);
904 			error = EPIPE;
905 			break;
906 		}
907 
908 		/*
909 		 * We have no more space and have something to offer,
910 		 * wake up select/poll/kq.
911 		 */
912 		if (space == 0) {
913 			wpipe->pipe_state |= PIPE_WANTW;
914 			++wpipe->pipe_wantwcnt;
915 			pipewakeup(wpipe, 1);
916 			if (wpipe->pipe_state & PIPE_WANTW)
917 				error = tsleep(wpipe, PCATCH, "pipewr", 0);
918 			++pipe_wblocked_count;
919 		}
920 		lwkt_reltoken(&wpipe->pipe_rlock);
921 
922 		/*
923 		 * Break out if we errored or the read side wants us to go
924 		 * away.
925 		 */
926 		if (error)
927 			break;
928 		if (wpipe->pipe_state & PIPE_WEOF) {
929 			error = EPIPE;
930 			break;
931 		}
932 	}
933 	pipe_end_uio(wpipe, &wpipe->pipe_wip);
934 
935 	/*
936 	 * If we have put any characters in the buffer, we wake up
937 	 * the reader.
938 	 *
939 	 * Both rlock and wlock are required to be able to modify pipe_state.
940 	 */
941 	if (wpipe->pipe_buffer.windex != wpipe->pipe_buffer.rindex) {
942 		if (wpipe->pipe_state & PIPE_WANTR) {
943 			lwkt_gettoken(&wpipe->pipe_rlock);
944 			if (wpipe->pipe_state & PIPE_WANTR) {
945 				wpipe->pipe_state &= ~PIPE_WANTR;
946 				lwkt_reltoken(&wpipe->pipe_rlock);
947 				wakeup(wpipe);
948 			} else {
949 				lwkt_reltoken(&wpipe->pipe_rlock);
950 			}
951 		}
952 		lwkt_gettoken(&wpipe->pipe_rlock);
953 		pipewakeup(wpipe, 1);
954 		lwkt_reltoken(&wpipe->pipe_rlock);
955 	}
956 
957 	/*
958 	 * Don't return EPIPE if I/O was successful
959 	 */
960 	if ((wpipe->pipe_buffer.rindex == wpipe->pipe_buffer.windex) &&
961 	    (uio->uio_resid == 0) &&
962 	    (error == EPIPE)) {
963 		error = 0;
964 	}
965 
966 	if (error == 0)
967 		vfs_timestamp(&wpipe->pipe_mtime);
968 
969 	/*
970 	 * We have something to offer,
971 	 * wake up select/poll/kq.
972 	 */
973 	/*space = wpipe->pipe_buffer.windex - wpipe->pipe_buffer.rindex;*/
974 	lwkt_reltoken(&wpipe->pipe_wlock);
975 	pipe_rel_mplock(&mpsave);
976 	return (error);
977 }
978 
979 /*
980  * MPALMOSTSAFE - acquires mplock
981  *
982  * we implement a very minimal set of ioctls for compatibility with sockets.
983  */
984 int
985 pipe_ioctl(struct file *fp, u_long cmd, caddr_t data,
986 	   struct ucred *cred, struct sysmsg *msg)
987 {
988 	struct pipe *mpipe;
989 	int error;
990 	int mpsave;
991 
992 	pipe_get_mplock(&mpsave);
993 	mpipe = (struct pipe *)fp->f_data;
994 
995 	lwkt_gettoken(&mpipe->pipe_rlock);
996 	lwkt_gettoken(&mpipe->pipe_wlock);
997 
998 	switch (cmd) {
999 	case FIOASYNC:
1000 		if (*(int *)data) {
1001 			mpipe->pipe_state |= PIPE_ASYNC;
1002 		} else {
1003 			mpipe->pipe_state &= ~PIPE_ASYNC;
1004 		}
1005 		error = 0;
1006 		break;
1007 	case FIONREAD:
1008 		*(int *)data = mpipe->pipe_buffer.windex -
1009 				mpipe->pipe_buffer.rindex;
1010 		error = 0;
1011 		break;
1012 	case FIOSETOWN:
1013 		get_mplock();
1014 		error = fsetown(*(int *)data, &mpipe->pipe_sigio);
1015 		rel_mplock();
1016 		break;
1017 	case FIOGETOWN:
1018 		*(int *)data = fgetown(mpipe->pipe_sigio);
1019 		error = 0;
1020 		break;
1021 	case TIOCSPGRP:
1022 		/* This is deprecated, FIOSETOWN should be used instead. */
1023 		get_mplock();
1024 		error = fsetown(-(*(int *)data), &mpipe->pipe_sigio);
1025 		rel_mplock();
1026 		break;
1027 
1028 	case TIOCGPGRP:
1029 		/* This is deprecated, FIOGETOWN should be used instead. */
1030 		*(int *)data = -fgetown(mpipe->pipe_sigio);
1031 		error = 0;
1032 		break;
1033 	default:
1034 		error = ENOTTY;
1035 		break;
1036 	}
1037 	lwkt_reltoken(&mpipe->pipe_wlock);
1038 	lwkt_reltoken(&mpipe->pipe_rlock);
1039 	pipe_rel_mplock(&mpsave);
1040 
1041 	return (error);
1042 }
1043 
1044 /*
1045  * MPSAFE
1046  */
1047 static int
1048 pipe_stat(struct file *fp, struct stat *ub, struct ucred *cred)
1049 {
1050 	struct pipe *pipe;
1051 	int mpsave;
1052 
1053 	pipe_get_mplock(&mpsave);
1054 	pipe = (struct pipe *)fp->f_data;
1055 
1056 	bzero((caddr_t)ub, sizeof(*ub));
1057 	ub->st_mode = S_IFIFO;
1058 	ub->st_blksize = pipe->pipe_buffer.size;
1059 	ub->st_size = pipe->pipe_buffer.windex - pipe->pipe_buffer.rindex;
1060 	ub->st_blocks = (ub->st_size + ub->st_blksize - 1) / ub->st_blksize;
1061 	ub->st_atimespec = pipe->pipe_atime;
1062 	ub->st_mtimespec = pipe->pipe_mtime;
1063 	ub->st_ctimespec = pipe->pipe_ctime;
1064 	/*
1065 	 * Left as 0: st_dev, st_ino, st_nlink, st_uid, st_gid, st_rdev,
1066 	 * st_flags, st_gen.
1067 	 * XXX (st_dev, st_ino) should be unique.
1068 	 */
1069 	pipe_rel_mplock(&mpsave);
1070 	return (0);
1071 }
1072 
1073 /*
1074  * MPALMOSTSAFE - acquires mplock
1075  */
1076 static int
1077 pipe_close(struct file *fp)
1078 {
1079 	struct pipe *cpipe;
1080 
1081 	get_mplock();
1082 	cpipe = (struct pipe *)fp->f_data;
1083 	fp->f_ops = &badfileops;
1084 	fp->f_data = NULL;
1085 	funsetown(cpipe->pipe_sigio);
1086 	pipeclose(cpipe);
1087 	rel_mplock();
1088 	return (0);
1089 }
1090 
1091 /*
1092  * Shutdown one or both directions of a full-duplex pipe.
1093  *
1094  * MPALMOSTSAFE - acquires mplock
1095  */
1096 static int
1097 pipe_shutdown(struct file *fp, int how)
1098 {
1099 	struct pipe *rpipe;
1100 	struct pipe *wpipe;
1101 	int error = EPIPE;
1102 	int mpsave;
1103 
1104 	pipe_get_mplock(&mpsave);
1105 	rpipe = (struct pipe *)fp->f_data;
1106 	wpipe = rpipe->pipe_peer;
1107 
1108 	/*
1109 	 * We modify pipe_state on both pipes, which means we need
1110 	 * all four tokens!
1111 	 */
1112 	lwkt_gettoken(&rpipe->pipe_rlock);
1113 	lwkt_gettoken(&rpipe->pipe_wlock);
1114 	lwkt_gettoken(&wpipe->pipe_rlock);
1115 	lwkt_gettoken(&wpipe->pipe_wlock);
1116 
1117 	switch(how) {
1118 	case SHUT_RDWR:
1119 	case SHUT_RD:
1120 		rpipe->pipe_state |= PIPE_REOF;		/* my reads */
1121 		rpipe->pipe_state |= PIPE_WEOF;		/* peer writes */
1122 		if (rpipe->pipe_state & PIPE_WANTR) {
1123 			rpipe->pipe_state &= ~PIPE_WANTR;
1124 			wakeup(rpipe);
1125 		}
1126 		if (rpipe->pipe_state & PIPE_WANTW) {
1127 			rpipe->pipe_state &= ~PIPE_WANTW;
1128 			wakeup(rpipe);
1129 		}
1130 		error = 0;
1131 		if (how == SHUT_RD)
1132 			break;
1133 		/* fall through */
1134 	case SHUT_WR:
1135 		wpipe->pipe_state |= PIPE_REOF;		/* peer reads */
1136 		wpipe->pipe_state |= PIPE_WEOF;		/* my writes */
1137 		if (wpipe->pipe_state & PIPE_WANTR) {
1138 			wpipe->pipe_state &= ~PIPE_WANTR;
1139 			wakeup(wpipe);
1140 		}
1141 		if (wpipe->pipe_state & PIPE_WANTW) {
1142 			wpipe->pipe_state &= ~PIPE_WANTW;
1143 			wakeup(wpipe);
1144 		}
1145 		error = 0;
1146 		break;
1147 	}
1148 	pipewakeup(rpipe, 1);
1149 	pipewakeup(wpipe, 1);
1150 
1151 	lwkt_reltoken(&wpipe->pipe_wlock);
1152 	lwkt_reltoken(&wpipe->pipe_rlock);
1153 	lwkt_reltoken(&rpipe->pipe_wlock);
1154 	lwkt_reltoken(&rpipe->pipe_rlock);
1155 
1156 	pipe_rel_mplock(&mpsave);
1157 	return (error);
1158 }
1159 
1160 static void
1161 pipe_free_kmem(struct pipe *cpipe)
1162 {
1163 	if (cpipe->pipe_buffer.buffer != NULL) {
1164 		if (cpipe->pipe_buffer.size > PIPE_SIZE)
1165 			atomic_subtract_int(&pipe_nbig, 1);
1166 		kmem_free(&kernel_map,
1167 			(vm_offset_t)cpipe->pipe_buffer.buffer,
1168 			cpipe->pipe_buffer.size);
1169 		cpipe->pipe_buffer.buffer = NULL;
1170 		cpipe->pipe_buffer.object = NULL;
1171 	}
1172 }
1173 
1174 /*
1175  * Close the pipe.  The slock must be held to interlock against simultanious
1176  * closes.  The rlock and wlock must be held to adjust the pipe_state.
1177  */
1178 static void
1179 pipeclose(struct pipe *cpipe)
1180 {
1181 	globaldata_t gd;
1182 	struct pipe *ppipe;
1183 
1184 	if (cpipe == NULL)
1185 		return;
1186 
1187 	/*
1188 	 * The slock may not have been allocated yet (close during
1189 	 * initialization)
1190 	 *
1191 	 * We need both the read and write tokens to modify pipe_state.
1192 	 */
1193 	if (cpipe->pipe_slock)
1194 		lockmgr(cpipe->pipe_slock, LK_EXCLUSIVE);
1195 	lwkt_gettoken(&cpipe->pipe_rlock);
1196 	lwkt_gettoken(&cpipe->pipe_wlock);
1197 
1198 	/*
1199 	 * Set our state, wakeup anyone waiting in select/poll/kq, and
1200 	 * wakeup anyone blocked on our pipe.
1201 	 */
1202 	cpipe->pipe_state |= PIPE_CLOSED | PIPE_REOF | PIPE_WEOF;
1203 	pipewakeup(cpipe, 1);
1204 	if (cpipe->pipe_state & (PIPE_WANTR | PIPE_WANTW)) {
1205 		cpipe->pipe_state &= ~(PIPE_WANTR | PIPE_WANTW);
1206 		wakeup(cpipe);
1207 	}
1208 
1209 	/*
1210 	 * Disconnect from peer.
1211 	 */
1212 	if ((ppipe = cpipe->pipe_peer) != NULL) {
1213 		lwkt_gettoken(&ppipe->pipe_rlock);
1214 		lwkt_gettoken(&ppipe->pipe_wlock);
1215 		ppipe->pipe_state |= PIPE_REOF | PIPE_WEOF;
1216 		pipewakeup(ppipe, 1);
1217 		if (ppipe->pipe_state & (PIPE_WANTR | PIPE_WANTW)) {
1218 			ppipe->pipe_state &= ~(PIPE_WANTR | PIPE_WANTW);
1219 			wakeup(ppipe);
1220 		}
1221 		if (SLIST_FIRST(&ppipe->pipe_kq.ki_note))
1222 			KNOTE(&ppipe->pipe_kq.ki_note, 0);
1223 		lwkt_reltoken(&ppipe->pipe_wlock);
1224 		lwkt_reltoken(&ppipe->pipe_rlock);
1225 	}
1226 
1227 	/*
1228 	 * If the peer is also closed we can free resources for both
1229 	 * sides, otherwise we leave our side intact to deal with any
1230 	 * races (since we only have the slock).
1231 	 */
1232 	if (ppipe && (ppipe->pipe_state & PIPE_CLOSED)) {
1233 		cpipe->pipe_peer = NULL;
1234 		ppipe->pipe_peer = NULL;
1235 		ppipe->pipe_slock = NULL;	/* we will free the slock */
1236 		pipeclose(ppipe);
1237 		ppipe = NULL;
1238 	}
1239 
1240 	lwkt_reltoken(&cpipe->pipe_wlock);
1241 	lwkt_reltoken(&cpipe->pipe_rlock);
1242 	if (cpipe->pipe_slock)
1243 		lockmgr(cpipe->pipe_slock, LK_RELEASE);
1244 
1245 	/*
1246 	 * If we disassociated from our peer we can free resources
1247 	 */
1248 	if (ppipe == NULL) {
1249 		gd = mycpu;
1250 		if (cpipe->pipe_slock) {
1251 			kfree(cpipe->pipe_slock, M_PIPE);
1252 			cpipe->pipe_slock = NULL;
1253 		}
1254 		if (gd->gd_pipeqcount >= pipe_maxcache ||
1255 		    cpipe->pipe_buffer.size != PIPE_SIZE
1256 		) {
1257 			pipe_free_kmem(cpipe);
1258 			kfree(cpipe, M_PIPE);
1259 		} else {
1260 			cpipe->pipe_state = 0;
1261 			cpipe->pipe_peer = gd->gd_pipeq;
1262 			gd->gd_pipeq = cpipe;
1263 			++gd->gd_pipeqcount;
1264 		}
1265 	}
1266 }
1267 
1268 /*
1269  * MPALMOSTSAFE - acquires mplock
1270  */
1271 static int
1272 pipe_kqfilter(struct file *fp, struct knote *kn)
1273 {
1274 	struct pipe *cpipe;
1275 
1276 	cpipe = (struct pipe *)kn->kn_fp->f_data;
1277 
1278 	switch (kn->kn_filter) {
1279 	case EVFILT_READ:
1280 		kn->kn_fop = &pipe_rfiltops;
1281 		break;
1282 	case EVFILT_WRITE:
1283 		kn->kn_fop = &pipe_wfiltops;
1284 		if (cpipe->pipe_peer == NULL) {
1285 			/* other end of pipe has been closed */
1286 			return (EPIPE);
1287 		}
1288 		break;
1289 	default:
1290 		return (EOPNOTSUPP);
1291 	}
1292 	kn->kn_hook = (caddr_t)cpipe;
1293 
1294 	knote_insert(&cpipe->pipe_kq.ki_note, kn);
1295 
1296 	return (0);
1297 }
1298 
1299 static void
1300 filt_pipedetach(struct knote *kn)
1301 {
1302 	struct pipe *cpipe = (struct pipe *)kn->kn_hook;
1303 
1304 	knote_remove(&cpipe->pipe_kq.ki_note, kn);
1305 }
1306 
1307 /*ARGSUSED*/
1308 static int
1309 filt_piperead(struct knote *kn, long hint)
1310 {
1311 	struct pipe *rpipe = (struct pipe *)kn->kn_fp->f_data;
1312 	int ready = 0;
1313 
1314 	lwkt_gettoken(&rpipe->pipe_rlock);
1315 	lwkt_gettoken(&rpipe->pipe_wlock);
1316 
1317 	kn->kn_data = rpipe->pipe_buffer.windex - rpipe->pipe_buffer.rindex;
1318 
1319 	/*
1320 	 * Only set EOF if all data has been exhausted
1321 	 */
1322 	if ((rpipe->pipe_state & PIPE_REOF) && kn->kn_data == 0) {
1323 		kn->kn_flags |= EV_EOF;
1324 		ready = 1;
1325 	}
1326 
1327 	lwkt_reltoken(&rpipe->pipe_wlock);
1328 	lwkt_reltoken(&rpipe->pipe_rlock);
1329 
1330 	if (!ready)
1331 		ready = kn->kn_data > 0;
1332 
1333 	return (ready);
1334 }
1335 
1336 /*ARGSUSED*/
1337 static int
1338 filt_pipewrite(struct knote *kn, long hint)
1339 {
1340 	struct pipe *rpipe = (struct pipe *)kn->kn_fp->f_data;
1341 	struct pipe *wpipe = rpipe->pipe_peer;
1342 	int ready = 0;
1343 
1344 	kn->kn_data = 0;
1345 	if (wpipe == NULL) {
1346 		kn->kn_flags |= EV_EOF;
1347 		return (1);
1348 	}
1349 
1350 	lwkt_gettoken(&wpipe->pipe_rlock);
1351 	lwkt_gettoken(&wpipe->pipe_wlock);
1352 
1353 	if (wpipe->pipe_state & PIPE_WEOF) {
1354 		kn->kn_flags |= EV_EOF;
1355 		ready = 1;
1356 	}
1357 
1358 	if (!ready)
1359 		kn->kn_data = wpipe->pipe_buffer.size -
1360 			      (wpipe->pipe_buffer.windex -
1361 			       wpipe->pipe_buffer.rindex);
1362 
1363 	lwkt_reltoken(&wpipe->pipe_wlock);
1364 	lwkt_reltoken(&wpipe->pipe_rlock);
1365 
1366 	if (!ready)
1367 		ready = kn->kn_data >= PIPE_BUF;
1368 
1369 	return (ready);
1370 }
1371